Good Enough + Everywhere: Google's Bundling Playbook
Google is using Microsoft's 1995 strategy to catch OpenAI. But this time might be different.
I canceled my Google One Pro subscription Saturday morning. By evening, I had resubscribed.
The reversal started when I looked at what I actually needed. I’m a perpetual AI tire-kicker—rotating through models the way some people sample streaming services, always chasing the next capability. I’d signed up for Google One Pro to access Gemini, tested it, and decided it was time to downgrade back to basic storage.
Then I did the math.
Two terabytes of cloud storage alone costs $9.99 a month. Google One bundles that same storage with Gemini for $19.99. ChatGPT Plus—costs $20 a month and includes no storage at all. For essentially the same price as ChatGPT alone, I could have both the storage I needed anyway and Google’s premium AI.
For the first time since ChatGPT launched, I found myself hovering over a different cancel button—the one for my OpenAI subscription.
That shift in thinking felt significant. Not because of my particular subscription choices, but because of what it revealed about how competition actually works in technology markets.
I’ve seen this pattern before. Thirty years ago, to be exact.
August 9, 1995
Trading on Netscape Communications stock was delayed for nearly two hours that morning. Order imbalances kept shares from hitting the NASDAQ floor—too much demand, not enough supply. When trading finally opened, the stock that had been priced at $28 the night before opened at $71.
By market close, Netscape was valued at $2.9 billion. The company was sixteen months old. It had yet to make a dollar in profit. Marc Andreessen, the twenty-four-year-old co-founder, appeared barefoot on the cover of Time magazine. His stake was worth $59 million.
The IPO created a new vocabulary. Investors started talking about “Netscape moments”—high-visibility public offerings that signal the dawn of a new industry. Netscape Navigator had captured over 80% of the browser market. The internet was arriving, and Netscape owned the gateway.
At Microsoft’s headquarters in Redmond, Washington, Bill Gates was paying attention.
Three months earlier—May 26, 1995—Gates had sent an internal memo to Microsoft executives. The subject line read “The Internet Tidal Wave.”
“The Internet is a tidal wave,” Gates wrote. “It changes the rules. It is an incredible opportunity as well as [an] incredible challenge.”
He continued: “I have gone through several stages of increasing my views of [the internet’s] importance. Now I assign the internet the highest level of importance. The Internet is the most important single development to come along since the IBM PC was introduced in 1981. It is even more important than the arrival of the graphical user interface.”
The telling detail came from Gates’ personal experience. After browsing the web for ten hours, he noted he “had not seen a single Word .DOC” or other Microsoft file formats. An entire digital world was emerging where Microsoft didn’t exist.
Netscape looked like the future. Microsoft looked late.
The question was whether Microsoft could catch up before Netscape turned its browser dominance into something larger—an entire platform for the emerging web. It was a race between innovation and distribution.
Twenty-seven years later, the same race would begin again.
November 30, 2022
ChatGPT reached one million users in five days. Instagram, the previous record holder, had taken seventy-five days to hit the same milestone. Within weeks, “ChatGPT” became synonymous with “AI” itself. The application threatened to unbundle search from Google—to do to Google what Netscape had threatened to do to Microsoft thirty years earlier.
At Google’s Mountain View headquarters, the response mirrored what had happened at Microsoft. CEO Sundar Pichai issued a “code red” in December 2022. He called Larry Page and Sergey Brin—the company’s co-founders who had stepped down more than three years earlier—back for emergency meetings.
Page and Brin attended several strategy sessions with executives, reviewed Google’s AI product roadmap, and approved plans to accelerate chatbot features in Search. Engineers worked through holidays. Sergey Brin, who hadn’t written code at Google in years, started programming again.
OpenAI’s valuation would eventually reach $500 billion. ChatGPT looked unstoppable.
Google looked late.
Both companies—Microsoft in 1995, Google in 2022—faced the same realization: an upstart was threatening their core business, and they might have already missed their window to respond.
The race was on. Could the incumbent gain innovation before the startup gained distribution?
The First Stumble
Microsoft’s initial response was underwhelming. Fifteen days after Netscape’s IPO, on August 24, 1995, Microsoft released Internet Explorer 1.0 as part of the Windows 95 Plus! add-on package. The reviews were brutal—Navigator commanded over 80% of the browser market and nobody took Microsoft’s entry seriously. Several months later came IE 1.5, which added basic table rendering and remained hopelessly behind Netscape’s capabilities.
Microsoft looked outmatched. But at least they’d avoided public humiliation.
Google’s first attempt went worse.
On February 8, 2023—just over two months after the code red—Google demonstrated Bard to the public. The company had rushed the chatbot to market, and it showed. In the promotional materials, Bard confidently claimed that the James Webb Space Telescope “took the very first pictures of a planet outside of our own solar system.”
Astronomers noticed immediately. NASA’s records showed the European Southern Observatory’s Very Large Telescope had captured the first exoplanet image in 2004—nearly two decades before JWST launched. The error wasn’t subtle ambiguity or a matter of interpretation. It was simply, publicly, demonstrably wrong.
The market’s response was swift. The next day, Alphabet’s market value dropped $100 billion—roughly the entire market capitalization of Ford, Delta, and American Airlines combined, vaporized in a single trading session. Reports surfaced that senior leadership had overruled risk assessments to rush Bard out the door, sacrificing accuracy for speed.
Both Microsoft and Google had stumbled. Both looked vulnerable. Both appeared to have confirmed the conventional wisdom: they’d missed their window. The pattern seemed clear—upstart innovation beats incumbent bureaucracy. Fast-moving startups disrupt slow-moving giants. Netscape would own browsers. OpenAI would own AI.
Except that’s not what happened.
It turned out the incumbents didn’t need to be better. They just needed to be good enough.
Good Enough + Everywhere
In August 1996, a year after Internet Explorer 1.0’s dismal debut, Microsoft released Internet Explorer 3.0. This version was different. It was bundled directly into Windows 95—not as an add-on, but as an integral component of the operating system.
IE 3.0 supported Netscape’s plugin technology, added frames and JavaScript (though Microsoft called it “JScript”), and introduced Cascading Style Sheets. More importantly, it was free and already installed on every Windows PC.
A year later, in September 1997, Microsoft held a launch party in San Francisco for Internet Explorer 4.0, featuring a ten-foot-tall letter “e” logo. The next morning, Netscape employees found the giant “e” on their front lawn.
Microsoft spent over $100 million per year on Internet Explorer and assigned over 1,000 people to the project. But the real power wasn’t in the development budget. As antitrust documents would later show, “Microsoft bound Internet Explorer to Windows with contractual and, later, technological shackles in order to ensure the prominent (and ultimately permanent) presence of Internet Explorer on every Windows user’s PC system.”
The strategy worked. IE’s market share climbed from less than 10% in 1996 to approximately 40% in 1998. By 2002, it would reach 95%.
Microsoft didn’t build the best browser. They built a good enough browser and made it inescapable.
Every Surface
Between 2024 and 2025, Google followed the same playbook—but they had far more surfaces to occupy than Microsoft ever did.
Microsoft had Windows and Office. Google had Android on billions of phones, Gmail handling the world’s email, Search answering billions of queries daily, Photos storing memories, Docs managing work, Maps guiding travel. Every Google product became a vector for Gemini.
The integration wasn’t subtle. Open your email, and Gemini offers to draft responses. Upload photos, and Gemini organizes them. Ask a question in Search, and Gemini answers before showing traditional results. The AI didn’t require installation or activation—it simply appeared wherever Google already lived.
About twice as many U.S. Android users engage with Gemini through the operating system than through the standalone Gemini app. You don’t download anything or create a new account. Gemini is already there.
Then Google made the same move Microsoft had made with Internet Explorer: they made the bundle inescapable.
In February 2024, Google “hid” its 200GB storage-only plan from the signup flow. The company’s Google One AI Premium plan—2TB storage plus Gemini Advanced for $19.99 a month—became the default option for users who needed cloud storage. You can’t add Gemini Advanced to a basic storage plan. If you want the AI features, you subscribe to the bundled plan.
The strategy created friction in the same way Microsoft had weaponized thirty years earlier. Internet Explorer came preinstalled on Windows. You could download Netscape, but that meant going somewhere else, finding the installer, running it. Most people never bothered.
ChatGPT requires the same extra effort—but with a twist that makes the friction even more effective.
In January 2026, Google emphasized what Gemini could do with your personal information. The AI can analyze your Gmail to understand your interests, search your Docs to find relevant context, organize your Photos based on what matters to you. You still have to turn on the feature, but Google already has access to everything. Enabling Gemini with your personal data is just activating a feature within the ecosystem you already trust.
If you want ChatGPT to do the same, you need to explicitly grant OpenAI—a third-party company—access to your Gmail, your Drive, your Calendar. That’s not flipping a switch on a feature. That’s opening your private data to an outside company.
The permission moat might be more powerful than the distribution advantage. You could download Netscape despite Internet Explorer being preinstalled. But you can’t easily replicate the decade of email, documents, and photos you’ve entrusted to Google. The data itself becomes the lock-in.
Distribution solved the visibility problem. But bundling a mediocre product everywhere just gets you ignored everywhere. The real question was whether Google could turn distribution into improvement fast enough to matter.
The Rearview Mirror
Both Microsoft and Google kept shipping even when their products weren’t great. They took the criticism, the bad reviews, the $100 billion market cap drops. They kept putting software in front of users.
Distribution gave them something valuable: usage. Millions of people clicked on Internet Explorer because it was already there. Millions of people tried Gemini because it appeared in their Gmail. Each interaction generated data. Bug reports came back. Engineers fixed problems. The products improved.
Microsoft’s market share climbed from less than 10% in 1996 to 40% in 1998. By August 1996, IE 3.0 had achieved feature parity with Netscape. A year later, IE 4.0 was actually better—faster, more standards-compliant, and more stable than the increasingly buggy Netscape Navigator 4.0.
The same pattern repeated thirty years later. Between October 2024 and November 2025, Gemini grew from 90 million to 650 million users—sevenfold in twelve months. ChatGPT grew from 200 million to 810 million in the same period—still leading, but the gap was closing. Market share told the story more clearly: ChatGPT’s dominance dropped from 87.2% to 68%; Gemini climbed from 5.4% to 18.2%.
Then, on November 18, 2025, Google released Gemini 3. The model topped ChatGPT on major benchmarks. In January 2026, Gemini 3 Pro achieved a 1501 Elo score on user preference rankings—the first model to cross the 1500 threshold.
Two weeks later, Sam Altman sent an internal memo to OpenAI staff: “Code Red.” The company’s highest priority level. Teams working on advertising, health agents, shopping assistants were reassigned within hours. Daily war room calls. An eight-week sprint to improve ChatGPT’s quality and speed.
Exactly three years after Sundar Pichai’s code red over ChatGPT, Sam Altman was declaring code red over Gemini.
The market noticed. The day after Gemini 3 launched, Google’s stock jumped 3%. By January 2026, the company briefly overtook Apple as the second-most valuable company, crossing $4 trillion in market cap for the first time. Google’s stock had gained 65% in 2025—its sharpest rally since 2009.
Internet Explorer took about two years to go from “good enough” to competitive with Netscape. Gemini did it in three years.
The playbook had worked.
But This Time Might Be Different
The bundling playbook worked for Microsoft. Netscape’s market share collapsed. By 1998, they open-sourced Navigator. AOL acquired what remained for $4.2 billion—a tenth of their peak valuation.
But OpenAI isn’t Netscape.
Unlike Netscape, which faced Microsoft alone, OpenAI has its own bundler. Microsoft embedded ChatGPT into Windows, Office 365, and GitHub. They paid $13 billion to ensure OpenAI wouldn’t be fighting distribution with innovation alone. This isn’t purely incumbent versus upstart.
And the technology itself hasn’t stabilized. In the browser wars, “good enough” meant rendering HTML and running JavaScript. Once IE reached parity in 1996, the bar stayed relatively fixed. AI capabilities are evolving monthly—reasoning, agents, multimodal understanding. Gemini crossed the 1500 Elo threshold in January 2026. Six months from now, that might be table stakes. Bundling works when “good enough” stays good enough. Can it work when the ceiling keeps rising?
The AI market might also support what the browser market couldn’t: multiple winners playing different games. I currently have three AI subscriptions—Google, ChatGPT, and Claude. Browsers were winner-take-all; you used one at a time. AI tools are serving different needs. Anthropic isn’t trying to be everywhere—they’re building tools developers love and letting those developers bring Claude to enterprises. That’s a different game than consumer bundling.
History rhymes, but the variables are different. We’re watching the bundling playbook execute in real-time. Google has deployed it successfully. But this time the startup has distribution, the technology is still accelerating, and the market structure might allow for specialization.
For now, I still have all three subscriptions. History hasn’t finished rhyming yet.
Related: